Cervical spine injuries
Neck injuries are a common problem among blunt trauma victims with more than 8,000,000 cases being seen annually in U.S. and Canadian Emergency Departments (ED) [1
]. While the majority of these cases represent soft tissue injuries, 30,000 patients suffer cervical spine fractures or dislocations and approximately 10,000 suffer spinal cord injury [2
]. There are no readily available national Canadian data on ED visits such as those provided by the U.S. National Hospital Ambulatory Medical Care Survey [1
]. The prevalence of potential neck injury can, however, be reasonably estimated for Canadian EDs. Extrapolation, on a population basis, from reliable U.S. figures [1
] suggests that 1.3 million potential neck injury patients are seen annually in Canada. Only 0.9% of these patients are found to have cervical spine fractures or dislocations, even less (0.5%) have a spinal cord injury [5
Cervical spine radiography
Current use of cervical spine radiography for alert and stable trauma patients is very inefficient and highly variable. Most U.S. patients undergo radiography regardless of their clinical findings. Some maintain that all trauma patients should undergo such radiography [2
]. This is mostly because of reports suggesting that clinical judgement alone is inadequate to predict injuries [6
]. The American College of Surgeons recommends cervical spine radiography for all trauma patients with injury above the clavicle [7
]. Indeed, a survey found that 97% of 125 U.S. trauma centres routinely order cervical spine radiography for all trauma patients [13
]. In contrast, emergency physicians encounter a larger number of patients with very minor injuries, and some American emergency physicians are more selective about their use of cervical spine radiography.
Although selective use of cervical spine radiography is more common in Canada, we have shown that there is very large variation among hospitals and physicians in the use of radiography [14
]. Universal cervical spine radiography has been considered inefficient by many authors who also note that the yield of this radiography for fracture or dislocation is very low [15
], with the proportion of positive radiography being less than 3% in most trauma series [8
]. Most authors suggest that radiography may not be required in alert patients with no pain or tenderness of the neck [4
]. The huge number of normal cervical spine radiographs performed adds to health care costs [37
] as well as to the burden of time and effort for emergency department staff, and Emergency Medical Services (EMS) who are expected to immobilize all patients before transport.
Cervical spine immobilization
Because of the potential for spinal injury, paramedics go to great lengths to protect the cervical spine of trauma patients. Consequently, regardless of the presence of neck symptoms, most trauma victims transported to hospital in ambulances are protected by such measures as a backboard, collar, and head immobilization devices [4
]. Not only is this often unnecessary, the potential for clinical adverse effects and discomfort from immobilization have been well documented. Chest straps used in immobilization have a marked pulmonary restrictive effect, even in healthy non-smokers [38
]. Immobilization on a board leads to progressively worse pain in the head, neck, and back area, often resulting in the necessity to radiograph an otherwise normal spine in the ED [39
Because trauma victims should be seen rapidly at the hospital, paramedics are given only 15-20 minutes to evaluate and treat them in the field before transport. Even for minor trauma victims, cervical spine immobilization takes at least five minutes to apply, or up to 30% of the allowed field time. Unlike minor trauma victims coming to the ED using their own means of transportation, who are most commonly sent immediately to the waiting room area, similar trauma victims immobilized and transported by paramedics can wait up to three hours until an ED stretcher becomes available, also holding up the EMS crew who then becomes unavailable for the next community emergency.
Once on an ED stretcher, it is not unusual for these patients to remain with full immobilization for several hours until c-spine radiographs or computed tomography can be performed and interpreted. As well, efforts to obtain satisfactory c-spine radiographs often require repeated attempts. This consumes valuable time for physicians, nurses, and radiology technicians and distracts them from other urgent responsibilities [15
]. In addition, this delay compounds the burden of our crowded Canadian EDs in an era when they are under unprecedented pressures [42
]. The median length of stay for a patient evaluated in the stretcher area is approximately eight to 12 hours, whereas similar minor trauma victims arriving without immobilization can be evaluated and discharged in less than four hours from the waiting room area.
Clinical decision rules
Without the support of widely accepted guidelines, paramedics are likely to continue to immobilize all minor trauma victims. Clinical decision (or prediction) rules help to reduce the uncertainty of medical decision-making by standardizing the collection and interpretation of clinical data [45
]. A decision rule is derived from original research and may be defined as a decision-making tool that incorporates three or more variables from the history, physical examination, or simple tests. These decision rules help clinicians with bedside diagnostic or therapeutic decisions. To fully develop a clinically effective rule is a lengthy process that involves separate studies to derive, prospectively validate, and finally implement the rule. The methodological standards for the derivation and validation of decision rules are well described [49
Implementation to demonstrate the true effect on patient care is the ultimate test of a decision rule [53
]. Unfortunately, many clinical decision rules are not prospectively assessed to determine their accuracy, reliability, clinical sensibility, or potential impact on practice. This evaluation is critical because many statistically derived rules or guidelines fail to perform well when tested in a new population [54
]. The reason for this performance failure may be statistical, i.e., overfitting or instability of the original derived model [57
], or may be due to differences in prevalence of disease or differences in how the decision rule is applied [58
]. Most decision rules are never used after derivation because they are not adequately tested in validation or implementation studies [60
Previous guideline studies for use by physicians
In order to identify subgroups of trauma patients who need not undergo cervical spine radiography, many studies have been conducted in the past 15 years by emergency physicians [11
], trauma surgeons [2
], and radiologists [10
]. Unfortunately, these studies suffer from great variability in design and none could be considered methodologically robust [67
]. Of note are the U.S. based NEXUS Criteria which have received prominent attention after the publication of a huge validation study incorporating more than 34,000 patients [26
]. Clinicians in Canada, however, have found several of the criteria to be poorly reproducible, namely "presence of intoxication" and "distracting painful injuries". Moreover, we recently attempted a retrospective validation of the NEXUS criteria based upon an existing database of 8,924 patients and found that the criteria missed ten of 148 clinically important injuries, yielding a sensitivity of only 93% [46
]. We also found poor performance of the NEXUS criteria in our phase II prospective validation study [70
]. We believe that the NEXUS criteria lack the accuracy and reliability to be useful for widespread clinical use.
Previous guideline studies for use by paramedics
The necessity to immobilize all victims of blunt traumatic injuries during ambulance transport remains controversial. Despite the absence of difference in the neurological outcomes of 454 patients with blunt spinal injuries transported by a U.S. EMS system with full immobilization and Kuala Lumpur, Malaysia with no immobilization [71
], most EMS systems continue to use back board, collar, and head immobilization during transport. We have been able to identify three original research papers that assessed the potential for paramedics to evaluate the c-spine in the field. Domeier conducted a large prospective cohort study evaluating selective immobilization by paramedics in 13,357 patients, 415 of which had cervical spine injuries [72
]. Paramedics did not immobilize 33 of the 415 patients with spine injuries, none of which sustained a spinal cord lesion. Stroh retrospectively reviewed the health records of 861 patients transported to a trauma-receiving hospital using a selective immobilization strategy, and subsequently discharged with the diagnosis of cervical spine injury [73
]. Five injuries were missed by their c-spine clearance protocol, one of which resulted in an adverse outcome. Muhr compared the immobilization rate in 293 patients before and 281 patients after the implementation of a selective spinal immobilization strategy, and found a 33% reduction in the rate of immobilization [74
]. All three papers used the selective immobilization strategy described in the NEXUS studies. In Canada, the Canadian C-Spine Rule (CCR) is currently used in the city of Calgary and the province of Nova Scotia (without formal safety evaluation). Most other Canadian EMS are awaiting further safety evaluation studies before implementing such a program.
Previous physician Canadian C-Spine Rule studies
The results of phase I, the derivation of the CCR, were published in JAMA
in October 2001 [75
]. This prospective cohort study involving 8, 924 stable, alert adult trauma patients was conducted in 10 large Canadian community and teaching hospitals (1996-1999). The ED physicians evaluated each patient for 20 standardized clinical findings and recorded these on a data sheet prior to radiography. Where feasible, a second physician conducted an independent interobserver assessment. Those variables found to be both reliable (kappa > 0.6) and strongly associated with the outcome measure (p < .05) were combined using recursive partitioning statistical techniques. The final model was formulated into a clinician-friendly algorithm, the Canadian C-Spine Rule (Figure ). The rule stratifies patients into high-, medium-, and low-risk groups and requires evaluation of active range of motion for those in the low-risk group. This rule was cross-validated on the derivation sample and was found to identify all 151 cases of clinically important cervical spine injuries with a sensitivity of 100% (95% CI 98-100). The rule also performed with a specificity of 42.5% and would have required radiography for only 58.2% of patients, a 23.9% relative reduction from the current ordering rate of 76.5%.
Figure 1 The Canadian C-Spine Rule. The Canadian C-Spine Rule for alert (Glasgow Coma Scale score 15) and stable trauma patients for whom cervical spine injury is a concern, including patients with either posterior neck pain with any blunt mechanism of injury (more ...)
The results of phase II, the validation of the CCR by physicians, were published in the New England Journal of Medicine
in 2003 [70
]. This prospective cohort study was conducted in nine large Canadian EDs (1999-2002) and enrolled 8,283 potential neck injury patients. More than 340 physicians explicitly and prospectively assessed patients for both the CCR and the NEXUS Criteria prior to diagnostic imaging and a second physician independently assessed some patients. The primary outcome, clinically important cervical spine injury, was defined as any fracture, dislocation, or ligamentous instability requiring internal fixation or treatment with a halo, brace, or rigid collar. The CCR was found to be highly sensitive for clinically important cervical spine injuries, identifying 161 of 162 cases. In the combined phases I and II, the rule would have identified 312 of 313 cervical spine injury cases, a sensitivity of 99.7% (95% CI 98-100). We also found the rule to be very reliable with a kappa value of 0.65. At the same time, our study found the NEXUS Criteria to have inadequate sensitivity, fair reliability, and very little potential to reduce use of radiography.
The potential impact on ED crowding was assessed by measuring the mean length of stay in the ED for patients without cervical spine injury and for whom reliable times were available. Patients who did not undergo radiography (N = 1,997) spent almost two hours less time in the ED (123.2 vs. 232.9 min; P < 0.001) than patients who did undergo radiography (N = 4,608).
We evaluated current physician practice, without the use of the CCR, by noting the number of cases where patients with cervical spine fractures were discharged from the ED without the fracture having been identified. This occurred 14 times during the study and nine of these cases were clinically important cervical spine injuries. All these patients returned due to ongoing pain or were recalled by the radiology department one or more days after the initial ED visit. Fortunately, no patient suffered an adverse outcome. In one of the nine clinically important cervical spine injury cases, no radiography was ordered during the initial visit. In another seven of the nine cases, physicians misread the radiographs as normal and the radiologists subsequently identified the error. In the ninth clinically important cervical spine injury case, the initial radiograph was actually normal.
Results from phase IIIa, which took place in 12 Canadian EDs from 2004 to 2006 (n = 11,824 patients) were recently published [76
]. Phase IIIa was a matched-pair cluster design trial which compared outcomes during 12-month 'before' and 'after' periods at six 'intervention' and six 'control' EDs, stratified by teaching or community hospital status. All alert, stable adults presenting after acute, blunt head or neck trauma were enrolled. Sites were randomly allocated to either intervention or control groups. During the intervention-site after-period, active strategies were employed to implement the CCR into practice, including education, policy, and 'on-line' reminders. Outcomes included cervical spine imaging rates and missed injuries. From the before to after periods, the cervical spine imaging rate had a relative reduction
of 12.8% at the six intervention sites from 61.7% to 53.3% (P = 0.01) but a relative increase
of 12.5% at the six control sites from 52.8% to 58.9% (P = 0.03); this difference between groups was significant (P < 0.001). There were no missed c-spine injuries at the intervention sites. We concluded that, despite low baseline cervical spine imaging ordering rates, active implementation of the CCR by physicians led to a significant decrease in use of cervical spine imaging without missed injuries or patient morbidity. Widespread use of the CCR for clinical clearance of the c-spine could lead to reduced health care costs and more efficient patient flow in busy EDs.
Validation of the CCR by paramedics
The validation of the CCR by paramedics took place between 2002 and 2006 in seven EMS systems distributed in three Canadian provinces [77
]. The study population consisted of consecutive alert, stable, and cooperative adults transported by ambulance to the local lead trauma hospital after sustaining acute blunt trauma with potential injury to the neck. These are patients for whom standard basic trauma life support (BTLS) protocols require immobilization. Patient assessment was made by primary care or advanced care paramedics, who were trained by means of a 2-hour web-based training session followed by a practical demonstration to assess the CCR and the component clinical variables in a uniform manner. The paramedics recorded their findings along with their interpretation of the decision rule itself on a data collection form prior to arrival at hospital. They followed standard procedures for immobilization of patients and did not use the CCR as the basis for the decision to immobilize. The primary outcome, acute cervical spine injury, was defined as any fracture, dislocation, or ligamentous instability demonstrated on radiography. A clinically important cervical spine injury was defined as any injury requiring internal fixation or treatment with a halo, brace, or rigid collar. All enrolled patients who did not have radiography received telephone follow-up and were classified as having no acute cervical spine injury if they met all the previously validated explicit criteria at 14 days [78
We enrolled 2,393 eligible patients in the study. 1,126 patients were not evaluated with radiography, and required telephone follow-up. We reached 788 (70.0%) of those patients, among which 682 were determined to not have sustained a cervical spine injury according to our validated proxy assessment tool. A total of 1,949 enrolled patients had complete outcome assessments. Twelve (0.6%) had a clinically important cervical-spine injury. In two cases, the investigators could perform an independent assessment of the rule based on the paramedic care report, but could not evaluate the paramedic assessment of the rule based on their study data collection sheet. The characteristics of the 444 patients without outcome assessments were similar to those with radiographic evaluation, but were less likely to be admitted to the hospital.
Paramedics conservatively misinterpreted the rule in 320 patients (16.4%), including 154 cases (7.9%) where "dangerous mechanism" was overcalled and 166 cases (8.5%) where paramedics did not evaluate neck rotation as required by the CCR. The CCR assessment for these patients was later categorized by the investigators as "indeterminate". Patient characteristics for these 320 patients where similar to those for which the rule was followed accurately, with the exception that none of the 320 patients had a cervical-spine injury. Paramedics did not attempt to evaluate neck rotation in any of the 12 patients with a clinically important injury. Excluding the indeterminate cases, the results yield a calculated sensitivity of 100% (95% CI 74-100%) and a specificity of 38% (95% CI 36-40). We performed secondary analyses involving all 1,949 patients to determine the potential effect of indeterminate cases when the rule was assessed by paramedics. When the rule was assumed to be positive for all indeterminate cases, the specificity was 32.4% (95% CI, 31 to 34), and when the rule was assumed to be negative for all indeterminate cases, the specificity was 46.6% (95% CI, 45 to 49). The sensitivity and negative predictive value remained the same since there were no cervical spine injuries among the indeterminate cases.
We assessed the reliability of paramedic interpretation of the rule among 155 paramedics by measuring the kappa coefficient for interobserver agreement for each element of the rule. The kappa value for the overall interpretation of the rule was 0.93 (95% CI, 0.87 to 0.99). In addition, agreement for the 8 individual components of the CCR was also very good, with kappa values ranging from 0.66 to 1.00.
The paramedics were asked to indicate on a five-point Likert scale how comfortable they would be in applying the CCR to this patient. The results were very supportive: Paramedics were "very uncomfortable" or "uncomfortable" applying the Canadian C-Spine Rule in 9.5% of cases; they were "comfortable" or "very comfortable" in 81.7% of cases.
We evaluated the potential impact of the rule on the number of necessary immobilizations. If paramedics were allowed to use the rule, 62.2% (95% CI, 60 to 64) of recruited patients would have required immobilization compared to the actual immobilization rate of 100%.
Rationale for the study
We have previously derived (phase I) [75
] and validated the CCR in physician (phase IIa) [70
], ED triage nurse (phase IIb) [79
] and in paramedic (phase IIc) [77
] groups. We recently demonstrated successful implementation of the CCR by physicians in multiple hospitals (phase IIIa) [76
], with a decrease in diagnostic imaging use by physicians and no adverse events. An implementation study using ED triage nurses is under way (phase IIIb). While we hope to demonstrate that ED triage nurses can safely remove patient's cervical immobilization devices, it would be significantly more valuable if we could empower the paramedics to selectively forego immobilization in the first place, and avoid great discomfort to patients. This is a practice already adopted by a number of U.S. and Canadian EMS services. We now hope to move the CCR project forward to the next level (phase IIIc) by carefully preparing paramedics to selectively immobilize the c-spine of very low-risk trauma patients who are alert and stable. Many decision rules in the past have not been widely adopted because of a failure to study implementation issues. We believe that this proposed safety evaluation study is an essential step towards the widespread implementation of the CCR by paramedics across Canada. If this evaluation study is successful, we can then plan wider dissemination of paramedic clearance in a future effectiveness trial. However, the current proposed study must demonstrate both safety and efficacy before dissemination can occur.